• No se han encontrado resultados

The other aspect of manufacturing upon which system or machine reliability has a major influence is product quality. All manufacturing processes are subject to a rigorous degree of control to ensure process accuracy and stability. This can include continuous process monitoring and the full measurement of some product parameters. In addition the use of statistical monitoring methods such as statistical process control (SPC) etc are often utilised to ensure the process remains “in control” and operating effectively. The SPC process will monitor the set parameters of the manufactured product and quantify if the process is “in control” and statistically normal or veering towards the control boundaries. An early indication of machine failure is the deviation of the product parameters. This can create anomalies in the recorded data which are often attributed to special causes.

The reliability monitoring feature of the TRAM method will allow the Tata engineers to identify the least reliable production systems. It can thus support the implementation of improvements similar to the descaling system upgrade and the overall reliability characteristics of the whole production process can be improved. The continual monitoring of all system’s performance will identify if the improvement has been successful or if further action is required. This feature will have similar benefits on product quality through decreasing the process instability in this manufacturing area.

The author constructed The TRAM method to be user friendly; this should ensure that the use of the model will be widespread at the Port Talbot plant. This widespread application can significantly reduce the manpower requirements for analysing stoppage issues. It is estimated that the effective implementation of this analysis model can immediately decrease engineer workload by several hours per week. There is a minimum of ten engineers at the Hot Strip Mill and these personnel are duplicated in the other manufacturing areas at the Port Talbot plant. Therefore this contribution to their working patterns can release a significant amount of hours which can be dedicated to more proactive approach such as continuous improvement activities.

10 Conclusions

Research has been performed into identifying a reliability analysis methodology suitable for repairable systems installed in a challenging manufacturing environment. This research has led to the development of the Tata Reliability Analysis Model (TRAM) method. This method, whilst not using unique reliability analysis methods, Is a novel approach to formatting standard reliability analysis models to analyse and monitor repairable systems deployed in a long term manufacturing scenario. This is a “common sense” approach to improving the condition of manufacturing assets (machinery) through long term monitoring and analysis.

The Contributions of this research to Tata Steel are:

• Developed new methodology for the Reliability Analysis of repairable systems • Utilised an innovative step of combining three Reliability Analysis methods as

complimentary activities

• Constructed an automated Reliability Analysis model which fulfils the project remit. In addition the model is capable of long term monitoring of system reliability

• Delivered the new Reliability Analysis method to Tata Steel. The Reliability Model is installed in the Port Talbot Technology Group with a direct link to the HSM database.

The implementation of this analysis model in the Hot Strip Mill at Port Talbot steel works has led to the following conclusions:

The failure data acquisition system at the Hot Strip Mill will allow the acquisition of all failure data relating to this manufacturing process. Up until this point this data was presented in a format which is not readily transferable to reliability analysis techniques. However an attempt at applying the analysis model to other sections of the Port Talbot plant has highlighted the differences in the failure data recording methods being used. This detail is being used to assist in developing a uniform failure data recording method which can be applied to all the manufacturing units.

In undertaking the review of research to support this work it was identified that there are no readily identifiable long-term applications of reliability modelling techniques suitable for repairable mechanical systems being applied within the world- wide manufacturing environment. One of the main reasons for this is the disparity of the

repairable systems under review and the range of operating conditions seen by these systems over a long-term manufacturing period. This means that most of the failure data sets produced are often not statistically significant, a factor which makes the failure data sets unsuitable for many analysis techniques. The TRAM method is a new measure which can be applied to such systems and has been engineered specifically to meets these requirements. This research has shown that the three level analysis approach used does work in these cases and will react to changes in system operation.

The derived reliability analysis method operates by applying the most widely used reliability analysis technique, the Power Law, to all the failure data sets under review. The calculated results obtained from the analysis are compared through the most appropriate reliability values, goodness of fit tests and trend testing. When indicated the additional breakdown of the failure data sets into annual segments allows the identification of the section of the failure data set which is not statistically significant using the two supporting analysis methods which operate simultaneously. The examination of the InMTBF allows the identification of the medium term reliability trends in the system, thereby identifying disparities in the systems data set. The specific measure introduced by this research, the TMTBF allows the identification of short-term trends by the analysis model. This can identify changes in operating and machine conditions that may have influenced the failure data set structure. This is a new and innovative approach introduced by this research that overcomes issue of the non- statistical significance of the failure data sets. The author believes that this issue has up to this point limited the application of reliability analysis to repairable systems. This research has thus increased the application of such approaches to the manufacturing environment.

This research has introduced the application of these analysis methods in an automated model to allow the feature of non- statistical significance to be used as a tool. This feature is an important new element introduced by this research. It represents a major contribution to the establishment and increased utilisation of effective reliability analysis tools. This feature can also identify the inconsistencies in any system’s manufacturing performance through reverse engineering the calculated reliability values one can trace the root failure causes and special circumstances which affect the operational performance of any system. This has important ramifications in an

engineering environment which is influenced by many operational parameters. The identification of operational controls which are detrimental to the process operation allows these parameters to be modified and construct a more process friendly operational control system. This could improve operating efficiencies and bring additional benefits in process stability and product quality.

The TRAM methods construction in the form of three Excel workbooks in a self- contained folder allows it to be easily transferable to any other manufacturing area. The analysis method is predominantly automated and it utilises advanced spreadsheet techniques to achieve this feature. The analysis method is user friendly and does not require specialist training to operate.

This analysis method has been tested with four years operational data from the Hot Strip Mill manufacturing area. The analysis has shown that changes in all systems operational status can be easily identified.

It has been established that the ability to perform a robust reliability analysis on any repairable system will be beneficial in the identification, construction and monitoring of any process upgrade. In addition the ability to identify trends in system reliability will facilitate a more efficient maintenance regime. This will enable engineers to be released for new manufacturing issues which could further enhance process efficiency and product quality.

There have been several papers withdrawn from this thesis; these are currently undergoing the review process at several Journals. The papers are:

A repairable mechanical system reliability assessment methodology applied in a steelmaking context.

R.J.Owen, S.Porretta, R.Grosvenorand P.Prickett

Submitted to Reliability Engineering & System Safety (August 2011) The reliability analysis of mechanical systems; Robert J Owen; Roger

Grosvenor; Steve Porretta, Paul Prickett. Submitted to Reliability Engineering & System Safety (January 2011)

Applying Reliability Assessment to Identify and Verify Process Improvements in a Hot Strip Steel Mill Descaling System.

R. Owen, R. Grosvenor, S Porretta and P. Prickett

Documento similar